Download Milestone7

BISC/CS303
Milestone 7
Due: March 19, 2008 at the start of class
(E-mail solutions to “BISC/CS303 Drop Box”)
Name:
Gene:
Task 1:
Construction of guide tree
Suppose that you have generated the following table of pairwise similarity scores for six
orthologous sequences from the following organisms: fruit fly (Drosophila), dog (Canis),
mosquito (Anopheles), puffer fish (Fugu), humans (Homo), and zebrafish (Danio)
Drosophila Anopheles
46
Drosophila
Anopheles
Canis
Fugu
Homo
Danio
Canis
39
34
Fugu
41
30
57
Homo
49
33
58
53
Danio
43
36
60
54
50
Using the unweighted pair group method with arithmetic mean (UPGMA), draw the
guide tree for the six sequences.
Task 2:
Generating pairwise similarity and pairwise distance scores
Download the Python program MSA.py from the course website:
http://cs.wellesley.edu/~cs303/assignments/M7/MSA.py
Study this program. The program contains 5 genomic sequences: seq1, seq2, seq3,
seq4, and seq5. The program also contains several helpful functions that have already
been implemented, including the following:


globalAlignmentScore(s1, s2) determines the score of the optimal
global alignment of two sequences, s1 and s2.
generateComparableRandomSequence(s) generates a random genomic
sequence of the same length as s and of the same expected nucleotide distribution
as s. For example, if s is a sequence of 582 nucleotides with GC content of 64%,
then the function will generate a random sequence of 582 nucleotides with an
expected GC content of 64%.
Another function, calculateDistanceScore(s1, s2), has not yet been
implemented. You will implement calculateDistanceScore(s1, s2) later in
this Task.
First, fill in the table below by calculating the optimal pairwise global alignment score for
5
each of the   = 10 possible pairwise combinations of sequences involving seq1,
 2
seq2, seq3, seq4, and seq5.
seq1
seq1
seq2
seq3
seq4
seq5
seq2
seq3
seq4
seq5
Each entry in the table above corresponds to a similarity score (specifically, the optimal
global alignment score) of two sequences. Higher similarity scores represent more similar
sequences, and lower similarity scores represent less similar sequences.
Now, you will calculate a distance score for each pair of sequences. A distance score is a
dissimilarity score, i.e., higher distance scores represent less similar (more distant)
sequences, and lower distance scores represent more similar (less distant) sequences.
A distance score, D, for two sequences s1 and s2 can be calculated as follows:
D = 100.0 * (-ln(Snorm))
where ln refers to the natural logarithm. The natural logarithm of a number x can be
calculated in Python as math.log(x). In the above equation, Snorm is defined as
Snorm = (Sglobal - Srand) / (Siden - Srand)
Sglobal is the optimal global alignment score of s1 and s2.
Siden is the average of the optimal global alignment score of s1 aligned with s1 and the
optimal global alignment score of s2 aligned with s2.
Srand is the average of 1000 optimal global alignment scores. Each of the 1000 global
alignment scores is calculated by aligning two randomly generated sequences, the first
being of the same length and same expected nucleotide composition as s1, and the
second being of the same length and same expected nucleotide composition as s2.
Fill in the function calculateDistanceScore(s1, s2) so that it calculates the
distance score as described above for two sequences s1 and s2.
5
Now, fill in the table below by calculating the distance score for each of the   = 10
 2
possible pairwise combinations of sequences involving seq1, seq2, seq3, seq4, and
seq5.
seq1
seq1
seq2
seq3
seq4
seq5
seq2
seq3
seq4
seq5
Distance scores are useful because the branch lengths in phylograms are often
determined by the distance of two sequences. Longer branch lengths are used for more
distant sequences and shorter branch lengths are used for less distant sequences.
Based on the first table (of similarity scores) in this Task, use UPGMA to draw a guide
tree for the five sequences: seq1, seq2, seq3, seq4, and seq5. Now, using the
information in the second table (of distance scores) in this Task, redraw the guide tree as
a phylogram so that branch lengths in the tree are indicative of the distance between
sequences or clades. The branch lengths in the phylogram need not be drawn perfectly to
scale, but it should be the case throughout the tree that two sequences or clades with a
larger distance score have longer branch lengths than two sequences or clades with a
smaller distance score.
When submitting this milestone, attach your Python program, MSA.py, as well as your
phylogram.
Task 3:
Creation of annotated multiple sequence alignment (MSA)
In this Task, you will perform a multiple sequence alignment (MSA) of the 10 putative
orthologs from diverse organisms that you found in the previous milestone.
To perform the multiple sequence alignment, you will be using the ClustalW program
available from the EBI (European Bioinformatics Institute):
http://www.ebi.ac.uk/clustalw/
Make sure that the sequences you submit to ClustalW are in a single text file in FASTA
format. When ClustalW produces a phylogram or cladogram, the OTUs are labeled with
the first characters in the comment line (i.e., >The text right here). So, it is useful to
modify the comment line in your FASTA-formatted sequences to reflect the organism
names instead of the accession numbers.
ClustalW should take only a few seconds to generate a MSA of your orthologous
sequences. When ClustalW finishes aligning your sequences, inspect the results.



Scroll down the results page to find the phylogeny. Click the buttons “Show as
Phylogram Tree” and “Show as Cladogram Tree” to toggle between phylogram
and cladogram versions of the phylogeny.
At the top of the results page, click on the “Start Jalview” button to open an
interactive display of the MSA.
One of the advantages of a MSA is that it can provide insight into various
properties of a family of proteins. When studying your MSA, if you find
portions of your sequences that do not align well, you may want to consider
removing these non-homologous regions, particularly if they are interfering with
optimal alignment within the protein family. This may be useful if your original
sequences differ radically in length, especially at their ends. If this is the case for
your dataset, please repeat the alignment without the tail sequences. Similarly, if
particularly divergent sequences from your set of orthologs are interfering with
optimal alignment, you may want to consider omitting these divergent sequences
from the alignment.
After studying the results of your MSA, answer the following questions.
Which pair of sequences (out of all pairs of sequences, not necessarily including your
yeast gene) is the most similar? Which pair of sequences is the most dissimilar?
How well does your yeast gene seem to fit your protein family?
Are there particular regions within your protein family that appear to be more highly
conserved? If so, why might some regions be more highly conserved? If not, why might
there be uniform conservation throughout a protein family?
When submitting this milestone, please include your actual multiple sequence alignment
(such as the alignment file or the output provided by Jalview). Also, please make sure
that each of your sequences is labeled with the organism’s real name (e.g., Arabidopsis,
puffer fish, etc.).
Task 4
Phylogenetic tree concepts
Prior to completing this task please read the attached perspectives article by Baum
et al. entitled “The Tree-Thinking Challenge”.
Phylogenetic trees represent hypotheses of the evolutionary relationships between taxa.
Molecular data has provided an invaluable supplement to pre-existing trait-based
phylogenies, and has greatly improved our phylogenetic knowledge. However care must
be taken when constructing phylogenetic trees based on molecular data.
Phylogenetic trees based on gene sequences do not always correspond exactly to the
phylogenetic trees showing the evolutionary relationships between species. What are two
reasons this might happen?
When using molecular data to construct phylogenetic trees, protein sequence data is
generally more useful than nucleotide sequence data for inferring evolutionary events that
occurred in the distant past. Why is this?
When choosing genes for use in a phylogenetic analysis, it is important that the genes be
orthologous, otherwise results may be erroneous. In the absence of direct experimental
data, how can you be sure that the gene sequences that you pick for phylogenetic analyses
are truly orthologous genes?
What would be the consequence(s) of using a gene in your phylogenetic analysis that is a
paralog that is no longer functional (called a pseudogene) instead of an ortholog?
Among the Bacteria and the Archaea, transfer of genetic material from the genome of one
species to another is not uncommon. For example, a fragment of DNA might be
transferred from the chromosome of Escherichia coli to the chromosome of Enterobacter
cloacae, since these bacteria are often present together in mammalian gastrointestinal
tracts. This is a process called horizontal gene transfer or lateral gene transfer
(transfer of genetic material from parent to offspring is called vertical gene transfer).
What would be the consequence of using a gene that has been laterally transferred
recently in a phylogenetic analysis that includes both the horizontal gene transfer donor
bacterium and the horizontal gene transfer recipient bacterium?
Task 5
Epidemiological case study of HIV virus phylogeny
Human Immunodeficiency Virus (HIV) is a virus with a single-stranded RNA genome
that is 9749 nucleotides long. Because RNA replication is highly error prone when
compared to DNA replication, the HIV virus is constantly mutating. Many of these
nucleotide changes result in non-functional viruses, but some produce viable viruses with
altered cell surface antigens. This represents a significant challenge to producing an
effective HIV vaccine.
The HIV genome has 9 ORFs that produce 15 proteins, which is accomplished via the
action of a protease encoded by the HIV genome. Because this HIV protease has a
mechanism of action that is distinct from human proteases, and because HIV protease
activity is crucial for virus replication, a mixture of drugs that specifically block the HIV
protease (protease inhibitor) and the HIV polymerase (reverse transcriptase) has been
reasonably effective at controlling HIV infections. The outside of the HIV virus is coated
with a glycoprotein called gp120. gp120 specifically binds to CD4, a human cell surface
protein. gp120-CD4 binding is a critical event for viral binding of and subsequent
infection of the host cell.
In the late 1980’s, eight patients of an HIV-positive dentist in Florida were diagnosed as
being HIV-positive. Though many of the patients had had invasive dental procedures
performed (root canals, tooth extractions), an investigation by the Centers for Disease
Control did not uncover systematic hygienic lapses that might account for infection of the
patients. Additionally, there were no obvious ways in which the dentist might have
deliberately infected his patients.
In an attempt to determine whether the eight HIV-positive patients were infected by the
dentist, researchers isolated viral RNA from blood samples from the dentist, the infected
patients, and HIV-positive individuals in the area who had had no contact with the
dentist. The investigators then amplified DNA copies of the genomic RNA sequences via
the polymerase chain reaction (P.C.R.) and determined the nucleotide sequence of pieces
of the HIV gp120 gene. This data was then used to determine how closely related the
dentist’s HIV virus strain was to that of his patients, the individual who was a sexual
contact of the dentist’s, and the HIV-positive individuals who had not had contact with
the dentist.
In this task you will use some of the gp120 protein sequences described above to
recapitulate the phylogenetic analysis performed in the case of the Florida dentist. To
begin, download the file “HIV_data_set.txt” from the course web site:
http://cs.wellesley.edu/~cs303/assignments/M7/HIV_data_set.txt
This file contains gp120 protein sequence data from the dentist, his infected patients, one
of the dentist’s HIV-positive sexual contacts, and local controls (other HIV-positive
individuals in the area who had no contact with the dentist).
Using ClustalW, perform a multiple sequence alignment on and build a phylogenetic tree
using the HIV gp120 sequences in the file provided. When submitting this milestone,
please include your phylogenetic guide tree.
Based on your phylogenetic tree, do you think it is likely that the Florida dentist infected
any of the eight patients? Are there any patients who were unlikely to have been infected
by the dentist? Why?
Why was it important for researchers to collect HIV samples from people who had no
contact with the HIV-positive dentist as well as an individual with whom the dentist had
had sexual contact?
Why did the researchers choose the HIV gp120 protein for their phylogenetic analysis? If
the researchers performed this analysis with another HIV protein, do you think their
results and conclusions would be similar? Why?
One criticism of the phylogenetic analysis is that HIV virus can mutate within an
individual, thereby producing many distinct HIV viruses with different gene sequences,
which could obscure the phylogenetic analysis. How might a researcher try to
compensate for this possibility in their experimental approach?